Kalon L. Kelley, Consultant
Casual carpooling currently functions in environments where, because there is an
HOV lane, riders and passengers join together to gain access to that lane. In this
article, technology is introduced that will allow casual carpooling to function in
areas without HOV lanes by providing an administrative system that records actual
carpooling behavior so that incentives other than access to an HOV lane can be
made available. This technology also addresses some of the current shortcomings
associated with casual carpooling such as personal safety, the “free rider” problem,
and the disincentive to maximize the number of passengers sharing a ride. The article
concludes with comments on funding enhanced carpool programs including a hypo-
thetical comparison of a program providing monetary incentives for carpooling to
the alternative construction of new HOV lanes.
Casual carpooling refers to the sharing of a ride with a driver and one or more
passengers, where the ridesharing between the individuals is not established in
advance but coordinated on the spot. While there may be a variety of motives for
ridesharing, the expression refers primarily to commuting situations where the
driver is incentivized to pick up passengers so as to allow for transit using a high
occupancy vehicle (HOV) lane. This type of carpooling has been around for quite
a number of years and is characterized by its informality and lack of governance.
Meeting sites tend to evolve where there is reasonable parking (for passengers
who may drive to the site and leave their cars) and proximity to major transporta-
tion corridors that provide HOV lanes. Casual carpooling has become a recognized
Journal of Public Transportation, Vol. 10, No. 4, 2007
cultural phenomena in places as diverse as Washington D.C., Pittsburgh, Houston,
and San Francisco.
Casual carpooling is often considered a win-win activity. Drivers get access to tran-
sit lanes that reduce the length of their commute; passengers get a free commute
to work. The community benefits by a reduction in vehicles on the road with its
panoply of benefits.
Technology for Carpools
Radio frequency identification devices (RFID) are now used in a variety of environ-
ments involving vehicles. An overview of toll collection systems is described by
Saranow (005), and the U.S. Border Patrol is now using automatic vehicle iden-
tification (AVI) technology to speed border crossings (U.S. Customs 006). While
these transponders identify a vehicle, that is all they do.
RFID devices are also used to identify people. For example, a major oil company
makes available small keychain devices that can be waved by a gas pump to pay for
a gasoline purchase; this technology is being extended to other retailers.
The proposal herein is to use a transponder that, physically comparable to the
dashboard-positioned toll collection transponders, has the additional capability
of reading and storing in memory individual RFID devices when they are passed
in proximity to the transponder. Then, when interrogated, the transponder trans-
mits its ID along with the individual IDs that were read and stored.
There is a one-time registration process where a potential driver registers with the
administrator and receives an enhanced transponder. An individual participating
in the program as a passenger will register and receive an individual RFID device.
The casual carpool begins when a driver enters his vehicle and turns on the tran-
sponder. At a collection point, any rider entering the vehicle passes his RFID device
by the transponder, which beeps to indicate that the device was read and then
records in memory the rider’s ID. As the vehicle has occasion to travel by a reading
station, the transponder will transmit its own ID along with IDs identifying the
occupants. The reading station will periodically transfer its collected information
to the central processing system. This allows for a historical record of carpooling
events (who, where, and when) to be created effortlessly. This is the kind of infor-
mation now manually collected by various carpooling systems to allow for reward-
ing participants, with the implication that a framework for incentivizing people to
form traditional carpools can apply as well to casual carpooling.
Identifying the transit corridors to be monitored is critical as it will not in general
be possible to have transponder readers on all highways and surface streets that
might be used for commuting. With multiple corridors and destinations, the
placement of readers would need to be done with care, in particular to avoid situa-
tions where commuters are motivated to use a less efficient route simply to ensure
that they pass through a transponder reader and get credit for their commute.
Establishing a casual carpooling collection point would be subject to the same
rationale as at present (density of drivers/riders, parking availability, public transit
options), and it would probably be necessary to have a way of distinguishing “reg-
istered” passengers (those with RFID devices) from those without,3 as a driver will
not get credit for providing a ride to the latter.
Two different contexts for casual carpooling exist. The traditional context is where
an HOV or toll lane exist, and where a carpool is granted access or reduced-fee
privileges on that lane. But casual carpools could exist in any transportation cor-
ridor if there was an incentive for drivers and passengers to participate; for ease of
reference these will be called non-HOV carpools.
The caution “not to accept a ride from a stranger” is taught at an early age. While
there are characteristics of casual carpooling that tend to ameliorate that concern
(random nature of pick-ups; the typical requirement for there to be at least two
passengers with a driver; and in some cases the restricted corridor, as for example,
crossing of the San Francisco Bay Bridge), it clearly is a concern with a significant
number of people in line choosing to decline an offered ride (Burris and Wynn
006) or reporting that the accountability of the third person was a “major factor
contributing to their sense of safety” (Gidal 004). That there are websites that
exist to report on wayward behavior of drivers and/or passengers is indicative of
the concern. While the success of casual carpooling demonstrates that its partici-
pants have by and large adjusted to the security issue, the unanswered question is
how many people presently do not participate but might choose this commuting
model if security could be enhanced?
When a driver invites a rider to enter her vehicle with a transponder and the rider
has an individual RFID device, both driver and rider can be comforted by know-
Journal of Public Transportation, Vol. 10, No. 4, 2007
ing that the other person’s identity is on file and that the ridesharing activity will
be recorded as the vehicle passes reading stations. In addition, the administrative
system could be designed so as to allow a commuter to provide anonymous feed-
back. For example, suppose a passenger is picked up by someone who drives dan-
gerously. The system could be set up so that later in the day the passenger could
access the Internet, identify himself, and indicate that on that morning’s commute
the driver was not recommended because of their driving. Anonymity would be
preserved as the passenger would not learn who the driver was. Similarly the driver
could record passenger evaluations, again anonymously.4 If the system administra-
tor established behavioral standards, participants with a pattern of inappropriate
behavior reports could be warned or excluded from participation.
Reporting of Carpool Events
Self-reporting of carpooling events when required to obtain incentives is problem-
atic. Underreporting happens due to laziness or forgetfulness, while overreport-
ing can be a deliberate attempt to gain undeserved rewards. With the proposed
technology, this reporting would be fully automated, with the only task required
being the waving of an RFID device by the transponder when entering the carpool
How Many Passengers?
The codified “rules of conduct” associated with casual carpooling serves to put
boundaries on what drivers and passengers do when sharing rides (e.g., limited
conversation and no cell phones). One of the constraints on drivers is not to pick
up more than the minimum number of passengers needed to qualify for use of
an HOV lane (with some exceptions such as late in the commute period when
drivers are infrequent). From the viewpoint of the participants, this is a perfectly
sound policy: it increases the liquidity of the system by “rewarding” more drivers
with passengers. But the flip side is that it encourages low-volume ridesharing and
hence more vehicles on the road.
In an HOV environment, it would be better to maintain liquidity with a dynamic
definition of a qualifying carpool. Suppose there was at the carpool collection
point a message board that defined the number of passengers needed to qualify
for HOV access as of this point in time. During peak periods this number could
be set high (4+) so as to provide an incentive for maximizing vehicle load. This
would also serve to bring into equilibrium drivers and passengers, as a driver who
regularly has difficulty finding a complement of passengers will have a diminishing
incentive to leave home alone in hopes of being able to form a carpool. In nonpeak
periods the number of passengers required to qualify might be reduced so as both
to maintain system liquidity and to provide an incentive for drivers and passengers
to shift to nonpeak-hour commuting.
In non-HOV carpools, both drivers and passengers can be incentivized to maximize
vehicle occupancy, either by increasing the rewards to drivers for more passengers
or, more likely, adjusting the program parameters so that rewards are available
only for vehicles with a higher minimum number of passengers. Rewards could
also be adjusted to favor nonpeak-hour commuting. There are endless permuta-
tions of reward structures with which to experiment. One of the more intriguing
observations was made in an Atlanta program (Clean Air Campaign 004) that
showed a very substantial persistence (64%) of alternative transportation use as
long as nine months to a year after the subsidies were eliminated. This suggests a
program design that provided financial incentives initially might not need to be
forever wedded to maintaining them (see also Beroldo 987).
Haphazard Ride Matching
In any casual carpooling arrangement, the ability to match rider destinations with
driver destinations will be imperfect. In some situations this may not matter much
(e.g., in San Francisco with crossing the Bay Bridge), but in others (Washington
D.C. with its multiple collection points and often multiple destinations from each
point) it may. In some meeting places riders self-segregate by destination, while
in others the procedure is for the driver to call out his destination and then to
have the first riders in line for whom that destination works step forward. When
the queue is fairly small, this could be an aggravation for a driver who is unable to
complete his carpool. Signage held by riders as to destination can also be useful.
Some proposals and experiments use cellular phone technology to improve this
“on-the-fly” matching function (NelsonNygaard Consulting Associates 006), but
there are not yet any successful implementations.
While building a database of actual commuting behavior has powerful implica-
tions for creation of traditional “match lists” and prearranged carpools, it can also
be used with casual carpooling. For example, a transponder reader and a message
board could be located at the pick-up site. When a vehicle pulls up, the transpon-
Journal of Public Transportation, Vol. 10, No. 4, 2007
der could be read and the destination displayed on the message board thereby
facilitating decision making by those waiting. Similarly, if the passengers passed
their RFID devices by a reader as they came into the pick-up area, there could be
a message board for drivers indicating destinations and number of people wait-
ing for that destination. Having this information available could be useful to both
drivers and passengers (with no passengers there would be no reason to stop, and
with too many heading to the same destination a passenger might anticipate an
unacceptable waiting period).
Enforceability of HOV Lane Restrictions
There is quite a literature on enforceability of HOV restrictions and considerable
diversity among communities and states in terms of the rigor with which the multi-
occupant requirement is enforced. That there have been efforts to recover Federal
funding for HOV lanes where enforcement is poor (National Transportation Library
993) attests to the problem. Ultimately it comes down to a law official peering
into the window of a vehicle and trying to count occupants, a fairly difficult task to
perform in the best of circumstances, and made problematic by reclining passengers,
small children, dirty or tinted windows, or inclement weather.
Devices to read vehicle transponders can be positioned anywhere along a transit
corridor and can be made part of the portable equipment of an enforcement offi-
cial. When a vehicle passes a reading station, it is interrogated, and if the passenger
count is inappropriate, a warning message or a signal to an enforcement official
could follow. Vehicles without transponders (or with the transponder turned off)
would be treated as violators. The system provides for a technical validation of
occupancy rather than depending on sporadic visual observation.5
Diffusion of Rewards
HOV lanes are established to encourage behavioral change that would increase
the number of occupants in a vehicle.6 To the extent that the lane becomes avail-
able to users who meet the occupancy requirement but for whom there has been
no decision to trade off convenience for sharing a ride, the system gets degraded
(Lehman et al. 994). For example, a family on vacation has probably not decided
to include an extra family member just to qualify for HOV access but may none-
theless enjoy this access. A mother driving her children to school has probably not
made a carpooling decision, while another mother who drives her neighbor’s chil-
dren along with her own may well have. The inability of the system to distinguish
among participants who should have access to the reward from those who should
not results in bizarre results such as the family breadwinner who commutes to
work accompanied by spouse (to qualify for HOV access), where after drop-off the
spouse then returns home in the vehicle only to repeat the process at the end of
the day. In this case the availability of the HOV lane has doubled the vehicle usage
on the highway, resulting in two round-trips rather than one.
Minimizing “free riders” (vehicles that meet occupancy requirements and gain
access to an HOV lane without any behavioral change to increase occupancy lev-
els) should follow from the eligibility requirements established during registration.
For example, a program might only register commuters who have an identified
place of employment, thereby eliminating most of the fortuitous eligibility (exam-
ples cited above). Although some people will benefit without having changed
their behavior (e.g., a neighborhood carpool that shifts to utilize a new HOV lane),
and some people might benefit even when there is a questionable net societal gain
(e.g., the bus rider who chooses to casually carpool because it is faster), a careful
definition of eligibility should limit this.
First and Last Miles
One of the least tractable problems with casual carpooling is the “first mile” and “last
mile”—the connections from home to the casual carpooling collection point and
the connections from the drop-off point to work (with the directionality reversed
on the way home). One approach to addressing this could be to adopt the transit
center (or “hub-and-spokes”) approach used in many municipal bus lines, wherein
a passenger boards a bus and goes to the transit center and then shifts to another
bus to complete the trip. While this would not fit all urban areas, in places where it
was feasible the logistics are not difficult. Drivers and passengers would be identified
both in the approach to the hub as well as when exiting, and any monetary reward
programs could reflect what was happening. Drivers could be rewarded both for
bringing passengers to the hub and for transporting a probably different group to
their general area of employment. Message boards at the hub could be updated in
real time to show how many vehicles were approaching (along with estimates as to
time of arrival) that were bound for any particular destination.
In a non-HOV environment, the possibilities for fine-tuning the operation of an
enhanced casual carpooling program are endless. If, for example, persons resid-
Journal of Public Transportation, Vol. 10, No. 4, 2007
ing in an area consistently failed to find rides, the formula for rewarding drivers
could be adjusted so as to increase the incentive for picking up passengers in that
area, and this could be done until a state of approximate equilibrium between
passengers and drivers was reached. Conceptually, this could be viewed as a mini
municipal bus system without fixed schedules and operating without exchange
of money between riders and drivers, but operating with enough frequency that
concerns about not getting a ride are minimal.
Extending the Reach of Casual Carpooling
As noted earlier, casual carpooling is, at present, closely tied to the incentive of
HOV lane access, although there may be some ancillary incentives such as elimina-
tion of tolls. But the basic incentive from the driver’s viewpoint is reduced transit
time, and for passengers it is some mixture of reduced transit time and the savings
associated with not having to drive. When access to an HOV lane is not available
as an incentive, if an alternative incentive (such as cash or awards) were provided,
it would free casual carpooling from its dependence on HOV lanes thereby greatly
extending the environments in which casual carpooling could be effective.
Numerous employer-based programs reward carpoolers with economic benefits
(free parking, monthly cash payments, and merchandise awards), and these pro-
grams do not depend on the existence of HOV lanes to make them work. Because
they depend on self-reporting of carpool activities, the rewards are usually mod-
est or temporary, and they usually work only within a company and not across a
Estimates of what it would cost to induce someone to commute using alternative
transportation (other than an SOV) can be made using historic trip reduction
tables (Victoria Transport Policy Institute 005) projected to the current year as
well as reports from actual employer-based programs (Clean Air Campaign 004;
National Transportation Library 993; Victoria Transport Policy Institute 006).
The tables distinguish environments that differ by degree of transit infrastruc-
ture present as well as costs for parking, but the financial incentive needed to
cause significant shifts from SOVs to alternative modalities is not large. Given the
persistence of carpooling behavior once started, a program might start with rela-
tively large incentives that are reduced over time. Endless opportunities exist for
experimentation by a program manager, and data as to the relationship between
amount of incentive and participation would accumulate over time.
With the ability to monitor actual carpooling events, employers can incentivize
their employees to carpool as an alternative to using an SOV, and the program
need not be limited (as is typically the case now) to programs where all the par-
ticipants are employed by the same company.
Further, the ability of the system to record carpooling events could help in moti-
vating Congress to extend the Commuter Benefit Program to add carpools to
the list of modalities (currently public transit or vanpools) that are eligible for tax
benefits. That they have not been included has reflected the difficulty of verifying
carpooling events, but in the proposed environment, the historical record is cap-
tured and available for audit.
But the largest source of funds should come from the Regional Transportation
Boards and state and federal agencies that have as their mandate the construc-
tion and operation of transportation systems. The old paradigm of simply building
more roads to accommodate ever-increasing traffic should no longer be the only
game in town.
To illustrate the point, in the Regional Transportation Plan for Santa Barbara
County (Parsons Brinckerhoff 006), the estimates for construction of adding
an HOV lane in both directions of a .8-mile section of Highway 0 (currently
two lanes in each direction) came to $58 million (006 dollars), and this did not
include the cost of additional infrastructure at freeway exits and entrances. After
completion of some operational improvements (now scheduled to begin in 008),
the plan estimates that 600 vehicles per peak hour would need to be eliminated to
improve traffic flow to a “D” level of service (the minimum level of service accept-
able in the Santa Barbara Congestion Management Program.)
If the number of riders (including the driver) to qualify for a financial incentive
was set at 4+, this reduction could be achieved by moving 600 SOV drivers into
qualifying carpools during each of two peak hours. If drivers were incentivized by
a payment of $0/day (to provide a carpool to Santa Barbara in the morning and
then back home in the evening), and riders were incentivized by a payment of $4/
day, the total cost for each peak hour would be 00 (drivers) X $0 + 600 (former
SOV drivers) X $4 = $4,400. As the program would operate for two peak-hour
periods per day and approximately 50 workdays/year, the annual cost would be
$4400 X X 50 = $. million. If operational costs were $.5 million/year, the total
Journal of Public Transportation, Vol. 10, No. 4, 2007
amount would be $.7 million. This is well under the $3. million estimated annual
maintenance cost of the proposed HOV lane.
The empirical justification for the incentive numbers used in this example is that
() they are well in excess of what the trip reduction tables (Victoria Transport
Policy Institute 005) projected to 007 from their base year suggest would be
required to cause the needed percentage of drivers to shift from an SOV to alter-
native transportation, and () they also exceed the incentives actually used in a
variety of carpooling programs (Transportation Demand Institute 997; Victoria
Transport Policy Institute 007).
The comparison to HOV maintenance costs is imperfect (since the HOV lanes
would provide benefits at nonpeak commuting hours, including the weekends
when summer vacation traffic can cause congestion comparable to that of the
workday week), and a more complete analysis would include projections for
future years.7 However, the difference in capital costs8 is so large that it should
cause planners to rethink their historical commitment to capacity building.
Over the years a number of proposals have addressed the automation of rideshar-
ing (Niles and Tolliver 99; Dickerson 004). While from a systems architecture
point of view the vision of dynamic carpool formation is fascinating, the complex-
ity attending these proposals has made their adoption, even as pilot programs,
problematic. In contrast, the technology proposed herein is simple to implement
and easy for commuters to use. It holds the promise of extending the proven
concept of casual carpooling to a larger regional audience. For regions with HOV
lanes, performance can be enhanced, while for regions without HOV lanes, more
traditional incentive-based programs become possible.
The computation of net reduction in vehicles due to carpooling is complicated
by the fact that there can be mode displacements, for example persons choosing
to carpool who might otherwise use a mass transit alternative (Beroldo 990).
Whether the driver also obtains a personal RFID might depend on whether the
vehicle could have different operators on different days. If this were not the case,
then the vehicle’s transponder could also serve to identify the driver, and an RFID
device for the driver would not be required.
Perhaps by signage, “This line for registered riders only.”
This could be a bit tricky with multiple anonymous passengers as any evaluation
by the driver would need to be attributed to the right passenger.
Some enforcement issues remain (e.g., a driver potentially “borrowing” individual
RFID devices without having the individuals present in the vehicle), but discussion
of this topic goes beyond this article.
Occasionally other objectives exist. For example, some venues encourage pur-
chase of hybrid vehicles by opening HOV lanes to these vehicles apart from occu-
The rate at which incentive program costs increase when traffic increases and
more drivers need to become carpoolers may exceed the rate at which HOV main-
tenance costs increase.
The capital cost to implement the carpooling program should be comfortably
less than percent of the $58 million to construct the HOV lane.
Beroldo, S. J. 987. Duration of carpool and vanpool usage by clients of rides. Trans-
portation Research Record 30.
Beroldo, S. J. 990. Casual carpooling in the San Francisco Bay Area. Transportation
Quarterly 44 (January 990).
Burris, Mark W., and Justin Winn. 006. Slugging in Houston: Casual carpool pas-
senger characteristics. Transportation Research Board Annual Meeting Paper
Clean Air Campaign. 004. http://www.ewire.com/display.cfm/Wire_ID/35.
Dickerson, Stephen L. 004. U.S. Patent No. 6,697,730.
Gidal, Jessica W. M. 004. Driving with the stranger. LoudPaperMag 4 ().
Lehman, Christopher K., Preston L. Schiller, and Kristin Pauly. 994. Re-thinking
HOV—high occupancy vehicle facilities and the public interest. Institute for
Transportation and the Environment. Chesapeake Bay Foundation.
Journal of Public Transportation, Vol. 10, No. 4, 2007
National Transportation Library. 993. A guidance manual for implementing effec-
tive employer-based travel demand management programs. DOT, p.6 ff. http://
NelsonNygaard Consulting Associates. 006. RideNow! Evaluation draft report.
Alameda County Congestion Management Agency (September).
Niles, John S., and Paul A. Toliver. 99. IVHS technology for improving ridesharing.
Proceedings of the 1992 Annual Meeting of IVHS America.
Parsons Brinckerhoff Quade and Douglas, Inc. 006. 101 in motion final report.
Santa Barbara County Association of Governments (July). www.0inmotion.
Saranow, Jennifer. 005. Some new reasons to use high-tech tolls. Wall St. Journal
(August 5), Section D.
Transportation Demand Institute of the Association for Commuter Transporta-
tion. 007. See case studies p. –70 (August). www.epa.gov/otaq/statere-
U. S. Customs. 006. Website describes the Secure Electronic Network for Travelers
Rapid Inspection (SENTRI). http://www.cbp.gov/xp/cgov/travel/frequent_
Victoria Transport Policy Institute. 005. Trip reduction tables. http://www.vtpi.
Victoria Transport Policy Institute. 006. Ridesharing (see case studies and exam-
Victoria Transport Policy Institute. 007. Commuter financial incentives. http://
About the Author
Kalon L. Kelley (email@example.com) is a retired computer software entre-
preneur and college teacher. A member of the Association for Commuter Trans-
portation, Dr. Kelley has published articles on computer science, linguistics, and
transportation. He has been active in recent years in exploring transportation
alternatives for Santa Barbara County, California.